Factors from residency applications associated with performance in emergency medicine: a retrospective study

The landscape of emergency medicine residency recruitment has undergone significant changes in recent years necessitating data-driven selection approaches. We utilized a multidimensional framework to clarify why certain screening measures predict success, hypothesizing that metrics measuring applied cognitive competence and non-cognitive professionalism differentially predict residency performance. This retrospective single-center study evaluated 133 residents graduating between 2018–2024. Applicant data included USMLE scores, core clerkships, Standardized Letters of Evaluation (SLOEs), interviews, red flags, and demographic variables. Resident performance was evaluated retrospectively using a five-point Likert scale by core leadership and Clinical Competency Committee faculty. The primary outcome was resident performance, dichotomized into high (score 3–4) or non-high (score 0–2) based on five-point Likert evaluations by four core faculty members evaluated with univariable logistic regressions and corroborated with a proportional odds ordinal logistic regression sensitivity analysis. Ninety residents (67.7%) were categorized as high performers. In univariable logistic regression, higher USMLE Step 2 scores (OR 1.40 per 10-point increase, 95% CI 1.04–1.93, p = 0.03) and younger age on entry (OR 0.84 per year, 95% CI 0.73–0.96, p = 0.02) were associated with high performance; both associations remained consistent in the ordinal analysis (Step 2: OR 1.45, p = 0.01; Age: OR 0.88, p = 0.03). A higher composite SLOE score had an unadjusted negative association with the primary binary outcome (OR 0.80; 95% CI 0.64–0.99; p = 0.05), which was attenuated to non-significance in the ordinal model (OR 0.85, 95% CI 0.71–1.02, p = 0.07). Application red flags demonstrated complete separation (all 3 residents with red flags were in the lower performance tier, Fisher's exact p = 0.03). USMLE Step 1 scores, core clerkships, interview scores, degree conferred (MD vs. DO), sex, and honor society membership were not significantly associated with performance. In this cohort, higher USMLE Step 2 scores demonstrated a modest positive association with subjective global performance ratings, while candidate red flags were associated with lower ratings. Conversely, historical composite SLOE scores and interview scores did not reliably differentiate higher global ratings. Given the limitations of retrospective subjective global ratings by program leadership and unadjusted univariable comparisons, these findings provide exploratory baseline insights for holistic review rather than definitive selection criteria.